Operating enterprise automotive growth across programmatic, measurement, and dealer demand

Context

Toyota and Lexus programs ran through a single programmatic operation: evergreen demand, seasonal and holiday pushes, model launches and refreshes, and dealer actions, planned, built, verified, and measured together. The funnel ran from upper-funnel national awareness to lower-funnel dealer demand.

The media and ad-serving stack covered DV360, CM360, Meta, YouTube, and PMP and custom programmatic deals, with third-party audiences and DMPs, Oracle audiences, IAS / DoubleVerify, and Nielsen for data and verification. Evergreen programs typically ran at approximately $15K–$30K per month; seasonal and model campaigns at approximately $50K–$100K per month, often over several months.

Constraint

Nothing was fundamentally broken. The challenge was finding the next improvement anyway.

• Mature baseline. Easy wins were limited. Improvement meant finding incremental gains inside an established media system, not repairing a failing account.
• Audience complexity. Model intenders, luxury-auto intenders, competitor conquest, owners and replacement-cycle audiences, lease-maturity proxies, third-party auto data, demographic and HHI overlays, and custom combinations.
• Limited first-party data. Strong audience performance had to be built without broad access to mature first-party audiences.
• Media quality. Certain campaigns saw invalid or fraudulent traffic approaching ~20%, which required aggressive inventory controls, verification, and exclusion strategy.
• Creative constraints. Automotive creative was tightly controlled, so gains had to come from format, length, CTA, sequencing, rotation, audience, and placement decisions.
• Fragmented decision visibility. No unified cross-channel reporting layer showed which audiences, themes, CTAs, formats, and campaigns were creating incremental value.

The job was not to rescue a broken media program. It was to build a system that could keep finding the next 5%.

My role and scope

I served as Associate Marketing Director and hands-on programmatic lead, responsible for campaign architecture, build, optimization, audience strategy, measurement, reporting, and client-facing performance work. Support staff and contractors assisted with trafficking, verification, and creative rotation; I personally built the campaigns or reviewed implementation in detail.

I personally built the DV360 campaigns, created the naming conventions and taxonomy, separated campaigns by device, audience, geography, model, and funnel stage when performance justified it, and rebuilt structures when incremental performance required more control.

Core Toyota and Lexus creative was supplied under strict client and brand controls. I informed media-driven creative recommendations and testing; I did not create brand creative.

Approach

Continuous optimization became the operating model, with five levers managed at line-item level:

• Audience strategy. In-market shoppers by model and segment, competitor conquest, current and lapsed owners, replacement-cycle and lease-maturity proxies, luxury-auto intenders, dealership-radius and geo audiences, low-value geo exclusions, video-engaged retargeting, and frequency-based sequencing.
• Media optimization. Audience exclusions, frequency caps, device strategy, geo segmentation, dayparting, bid strategy, pacing, budget reallocation, inventory and app blocking, PMP and curated inventory, sequential messaging, and dealer-level targeting.
• Media quality. IAS / DoubleVerify controls, Nielsen and other validation, invalid-traffic monitoring, brand-safety controls, and inventory QA.
• Measurement and BI. CM360 / Floodlight implementation and QA, custom reporting, a cross-channel naming taxonomy, Python and JavaScript scripting where useful, Power BI / Google BI tooling, and manual QA where automated reporting could not be trusted.
• Creative learning, where permitted. CTA length, video length, bumper / 15s / 30s formats, skippable vs. non-skippable, placement, sequencing, and audience-creative combinations. Psychological buckets and consistent naming made creative performance comparable across channels.

Optimization and business outcomes fed back into audience and architecture decisions.

Outcomes

Across optimization periods, the same patterns kept returning:

These are directional ranges across different campaigns and windows, not one controlled test, and they vary by campaign and period.

Patterns that repeated: conquest audiences generally outperformed broad auto-intender audiences; model-specific audiences generally outperformed broad automotive-interest audiences; and device, geo, and daypart splits repeatedly surfaced incremental opportunities. Lower-funnel signals measured included dealer-locator actions, lead forms, credit applications, appointment and test-drive requests, and retargeting conversion behavior.

What changed

The advantage was not one optimization. It was the discipline to keep finding the next one.

• Media quality as performance. Fraud, viewability, inventory quality, and brand safety were treated as economic variables, not only compliance requirements.
• Audience intelligence. Broad automotive targeting evolved into increasingly specific combinations of model interest, conquest, lifecycle, demographic, behavioral, and geographic signals.
• Measurement discipline. Campaign, verification, Floodlight, site, and business-action data were brought into a common decision framework rather than optimized from platform dashboards alone.
• Operational leverage. Naming conventions, BI, scripts, and repeatable processes made it possible to read performance across millions of dollars in media without losing line-item-level control.